Tech Meridian ← LIVE FEED
PROMY MERIDIAN RU

NEWS · COMPANIES · #3826

Booking.com engineer outlines 20-year ML platform evolution, cites 400B predictions/day and new agentic AI foundation

At QCon London, Jabez Eliezer Manuel (Principal Engineer, Booking.com) reviewed two decades of the company's data-management and ML platform work, covering MySQL scaling and replication (managed with Orchestrator), the sunset of Hadoop and cloud migration, high‑throughput model serving and feature stores handling about 400 billion predictions per day, experimentation practices, personalization, and a newly described agentic AI foundation for long‑term memory. The presentation focused on platform engineering lessons and a deep dive into one platform's evolution rather than releasing new products or detailed specifications.

KEY POINTS

  1. At QCon London, Jabez Eliezer Manuel (Principal Engineer, Booking.com) reviewed two decades of the company's data-management and ML platform work, covering MySQL scaling and replication (managed with Orchestrator), the sunset of Hadoop and cloud migration, high‑throughput model serving and feature stores handling about 400 billion predictions per day, experimentation practices, personalization, and a newly described agentic AI foundation for long‑term memory.
  2. The presentation focused on platform engineering lessons and a deep dive into one platform's evolution rather than releasing new products or detailed specifications.
  3. The talk reveals how a major travel platform operates ML at extreme scale and signals enterprise interest in agentic AI and long‑term memory infrastructure.

WHY IT MATTERS

The talk reveals how a major travel platform operates ML at extreme scale and signals enterprise interest in agentic AI and long‑term memory infrastructure.

SOURCES & TIMELINE

1